Active contour and texture features hybrid model for breast cancer detection from ultrasonic images

Breast ultrasound is commonly used in the early detection of breast cancer. The existing geodesic-based methods use pre-defined filters that necessitate extensive prior knowledge to achieve the region of interest in input image. Furthermore, the majority of ultrasound images suffer from noise and acoustic shadowing, which reduces the accuracy of tumor detection. To make the breast ultrasound image more informative, the discriminative features can also be extracted to improve detection accuracy. This article proposes a method to combine Active Contour and Texture Feature Vectors to find discriminative patterns. A comprehensive set of discriminative features for cancer detection in ultrasound images is created by combining the two learning models. The Breast Ultrasound Images dataset is used to evaluate the suggested method and compare it to recently created algorithms. Experimental results reveal that the proposed approach outperforms the existing algorithms in terms of accuracy, recall, precision, Jaccard index, and F1 score.